No smartwatch or smart ring currently on the market can reliably measure your blood glucose without piercing the skin. The FDA issued a safety warning in February 2024 stating it has not authorized, cleared, or approved any such device. But the underlying technologies being developed and tested are real, and several competing approaches are racing to solve this problem. Here’s how they work and why it’s proven so difficult.
Light-Based Glucose Detection
The most common approach uses light to detect glucose molecules through your skin. A sensor on the back of the watch shines near-infrared light into your wrist tissue, then analyzes what comes back. Glucose molecules interact with light in predictable ways: they absorb, scatter, and rotate it differently than other molecules in your blood and tissue. By reading those patterns, the sensor attempts to calculate how much glucose is present.
One promising version of this uses a technique called Raman spectroscopy. When near-infrared light hits glucose molecules, a tiny fraction of that light scatters at wavelengths unique to glucose, almost like a molecular fingerprint. This specificity is a major advantage, because water (which makes up most of your tissue) doesn’t interfere much with the signal. The catch is that the light only penetrates less than one millimeter into your skin, meaning it’s sampling the outermost layers of tissue and the fluid between cells rather than blood directly.
Another light-based method fires short laser pulses into the skin. When glucose molecules absorb this light energy, they heat up slightly and expand, producing tiny ultrasonic waves. A microphone or vibration sensor on the watch picks up those waves, and the strength of the signal correlates with glucose concentration. Because sound waves travel through tissue with less scattering than light, this approach can potentially reach deeper. But the signal is extremely weak and requires sophisticated amplification and noise filtering.
Electrical Impedance Sensing
Instead of light, some devices send a tiny electrical current through the skin and measure how the tissue resists it. Glucose changes the electrical properties of the fluid surrounding your cells. Specifically, it alters how well tissue conducts electricity and how much electrical charge it can store (its capacitance). By measuring these properties across a range of frequencies, a sensor can isolate the signal that tracks with glucose levels.
Research has shown that the capacitance reading from skin tissue is the primary carrier of glucose information in this approach. However, the electrical signal is also affected by how hydrated your skin is. Studies using this method found that the sensor needed to distinguish between two overlapping effects: actual glucose changes and a separate “equilibration process” related to skin moisture. Separating these two signals requires complex statistical modeling, and even small errors in that separation can throw off the glucose reading.
Radio Frequency Sensing
A newer approach uses radio waves, similar in concept to RFID tags, to penetrate the skin and detect glucose. The idea is that different concentrations of glucose in your body’s fluids create distinct radio frequency signatures. A sensor beams radio waves into the tissue, and the way those waves are absorbed or reflected changes depending on how much glucose is present. Early proof-of-concept studies have shown this can distinguish between different glucose concentrations in controlled lab solutions, but translating that to the complex environment of a living human wrist is a far larger challenge.
Why Accuracy Remains the Core Problem
The fundamental difficulty is that glucose makes up a tiny fraction of what’s in your blood, and every non-invasive method must detect it through layers of skin, fat, and other tissue that vary enormously from person to person and even hour to hour. Several biological and environmental factors create noise that can overwhelm the glucose signal.
Temperature is one. Skin temperature on your wrist can range from about 20°C to 40°C depending on the environment and physical activity, and sensor performance changes across that range. In enzyme-based sweat sensors, for example, higher temperatures boost the signal while lower temperatures suppress it. Skin pH varies by roughly two units between individuals during exercise, which also shifts sensor readings significantly. Motion introduces its own artifacts: every time you move your wrist, the contact between the sensor and your skin shifts, distorting optical and electrical measurements.
Then there’s a built-in time delay. Most non-invasive approaches measure glucose in the fluid between cells rather than in blood directly. Glucose takes about 7 to 8 minutes to move from your bloodstream into this fluid, with all subjects in one study showing detectable levels within 10 minutes. That lag means a wrist sensor is always showing you where your glucose was several minutes ago, not where it is right now. For someone managing diabetes with insulin, even a few minutes of delay on top of sensor inaccuracy could lead to dangerous dosing errors.
Calibration: The Finger-Prick Problem
Even the best non-invasive prototypes can’t work entirely on their own. Because skin thickness, pigmentation, hydration, and blood flow vary so much between people, each device needs to be calibrated against a known-accurate glucose measurement. Early prototypes required a dozen or more finger-prick calibrations to build a personalized model, which largely defeats the purpose of a needle-free device.
More recent research has shown progress. A 2025 study published in Nature found that a light-based wearable could produce accurate readings for up to 30 days after a single calibration test, with no clinically unsafe results during that window. A once-monthly calibration is far more practical than daily finger pricks, but it still means the device isn’t truly standalone.
How Accurate Devices Need to Be
Accuracy in glucose monitoring is measured by something called MARD, the mean absolute relative difference between the sensor’s reading and a lab-grade reference. Current FDA-approved continuous glucose monitors that use a tiny needle under the skin achieve a MARD around 11 to 12%, meaning their readings are off by that percentage on average. To be clinically useful, a non-invasive watch would need to match or beat that number consistently.
No non-invasive wrist device has publicly demonstrated that level of accuracy across diverse populations and real-world conditions. The products you can buy today on sites like Amazon that claim to measure blood glucose use unvalidated technology, and the FDA has specifically warned that relying on them for diabetes management could lead to wrong medication doses, dangerously low blood sugar, confusion, coma, or death.
What Current Smartwatches Actually Measure
Mainstream smartwatches from companies like Apple, Samsung, and Garmin do not measure blood glucose. What they do measure, heart rate, blood oxygen, and skin temperature, uses some of the same optical sensor hardware that could theoretically support glucose detection. Apple and Samsung have both filed patents and are widely reported to be developing glucose features, but neither has released one.
The watches and rings currently sold with glucose claims are typically from lesser-known brands and have no clinical validation. They may display a number that looks like a glucose reading, but there is no evidence that number reflects your actual blood sugar. If you use insulin or other glucose-lowering medications, acting on a false reading from one of these devices could be life-threatening.